hts.plate package

Submodules

hts.plate.plate module

synopsis

The Plate Class.

class hts.plate.plate.Plate(data, name, **kwargs)

Bases: object

Plate describes all information connected to the readout_dict of a high throughput screen. This could be either several readouts of a plate, or the same plate across several plates.

name

Name of the plate

Type

str

width

Width of the plate

Type

int

height

Height of the plate

Type

int

KNOWN_DATA_TYPES[i]

The data associated to this Plate, e.g. a plate layout, or readouts.

Type

subclass of plate_data.PlateData

__str__()

Create string for Plate instance.

add_data(data_type, data, force=False, tag=None)

Add data of data_type to self.config_data

Add data of data_type to self.config_data

calculate_control_normalized_signal(data_tag_readout, negative_control_key, positive_control_key, data_tag_normalized_readout=None, local=True, **kwargs)

Normalize the signal in data_tag_readout, normalized by negative_control_key and positive_control_key.

Normalize the signal in data_tag_readout, normalized by negative_control_key and positive_control_key.

The normalization is calculated as: .. math:

y’ =

rac{y - mu_{nc}}{| mu_{nc} - mu_{pc}| }

For local==True, $mu_{nc}$ and $mu_{pc}$ are predicted locally to the well (using Gaussian processes). For local==False, $mu_{nc}$ and $mu_{pc}$ are estimated by the average control values across the plate.

Args:

data_tag_readout (str): The key for self.readout.data where the readouts are stored. negative_control_key (str): The name of the negative control in the plate layout. positive_control_key (str): The name of the positive control in the plate layout. data_tag_normalized_readout (str): The key for self.readout.data where the normalized readouts will be stored. local (Bool): If True, use Gaussian processes to locally predict the control distributions. Else, use

plate-wise control distributions.

calculate_linearly_normalized_signal(unnormalized_key, normalized_0, normalized_1, normalized_key)

Linearly normalize the data

\[\]

normalized__i =

rac{ x_{unnormalized_i} - hat{x_{low}} } { hat{x_{high}} - hat{x_{low}} }

normalized_0 are all wells (according to the plate layout) with mean(wells)==0 after normalization. normalized_1 are all wells (according to the plate layout) with mean(wells)==1 for normalization.

Args:

unnormalized_key (str): The key for self.readout.data where the unnormalized Readout instance is stored. normalized_key (str): The key for self.readout.data where the resulting normalized Readout instance will be stored. x_low (list of str): The list of names of all low fixtures in the plate layout (self.plate_data). x_high (list of str): The list of names of the high fixture in the plate layout (self.plate_data).

calculate_local_ssmd(data_tag_mean_pos, data_tag_mean_neg, data_tag_std_pos, data_tag_std_neg, data_tag_ssmd, **kwargs)

Calculate local SSMD values.

Parameters

• data_tag_mean_pos –

• data_tag_mean_neg –

• data_tag_std_pos –

• data_tag_std_neg –

• data_tag_ssmd –

Returns:

calculate_net_fret(donor_channel, acceptor_channel, fluorophore_donor='fluorophore_donor', fluorophore_acceptor='fluorophore_acceptor', buffer='buffer', net_fret_key='net_fret')

Calculate the net FRET signal for a donor acceptor FRET setup.

Calculate the net FRET signal for a donor acceptor FRET setup. Typical donor -> aceptor pairs include

• 414nm CFP -> 475nm -> YFP 525nm

• EU -> 615nm -> APC 665nm

The following wells are needed, for both channels

• donor Donor_fluorophor blank

• acceptor Acceptor_fluorophor blank

• buffer Buffer blank

The proportionality factor for donor compensation is then calculation as .. math:

p =

rac{hat{donor_{acceptor_channel}} - hat{buffer_{acceptor_channel}}}{hat{donor_{donor_channel}} - hat{buffer_{donor_channel}}}

Further, the net FRET signal f for all wells x may be calculated as: .. math:

netfret = x_{acceptor_channel} - hat{acceptor_{acceptor_channel}} - p cdot (x_{donor_channel} - hat{buffer_{donor_channel}})

Args:

donor_channel (str): The key for self.readout.data where the donor_channel Readout instance is stored. acceptor_channel (str): The key for self.readout.data where the acceptor_channel Readout instance is stored. fluorophore_donor (str): The name of the donor fluorophore in self.plate_data. fluorophore_acceptor (str): The name of the acceptor fluorophore in self.plate_data. buffer (str): The name of the buffer in self.plate_data. net_fret_key (str): The key for self.readout.data where the resulting net fret Readout instance will be stored.

calculate_normalization_by_division(unnormalized_key, normalizer_key, normalized_key)

The normalize data set is equal to a division of all data by the mean of a subset of the data. :param unnormalized_key: The key for self.readout.data where the unnormalized Readout instance is stored. :type unnormalized_key: str :param normalized_key: The key for self.readout.data where the resulting normalized Readout instance will be stored. :type normalized_key: str

calculate_significance_compared_to_null_distribution(data_tag_readout, sample_tag_null_distribution, data_tag_standard_score, data_tag_p_value, is_higher_value_better, **kwargs)

Calculate the standard score and p-value for all data (in data_tag_readout) compared to the null distribution defined by all data of sample_tag_null_distribution in data_tag_readout. Save as readouts with tags data_tag_standard_score and data_tag_p_value.

Assume that the samples in sample_tag_null_distribution follows a Gaussian distribution.

WARNING! For pvalue calculation, we assume that the control, which has lower mean values, is also supposed to show lower mean values. [Otherwise, we would have to introduce a boolean “pos_control_lower_than_neg_control.”]

Parameters

• data_tag_readout (str) – The key for self.readout.data where the readouts are stored.

• sample_tag_null_distribution (str) – The sample key (defined in plate layout) defining what sample will make up the null distribution that we compare all other samples to.

• data_tag_standard_score (str) – The key for self.readout.data where the standard scores will be stored.

• data_tag_p_value (str) – The key for self.readout.data where the p-values will be stored.

• **kwargs –

Returns:

classify_by_cutoff(data_tag_readout, data_tag_classified_readout, threshold, is_higher_value_better=True, is_twosided=False, **kwargs)

Map a dataset of float values to either binary (is_twosided==False) or [-1,0,1] (is_twosided==True), depending on whether values fall below threshold

Parameters

• data_tag_readout – The key for self.readout.data where the readouts are stored.

• data_tag_classified_readout – The key for self.readout.data where the True/False classification values will be stored.

• threshold –

Returns:

create(name=None, **kwargs)

Create Plate instance.

Create Plate instance.

Parameters

• path (str) – Path to input file or directory

• format (str) – Format of the input file, at current not specified

cross_validate_predictions(data_tag_readout, sample_tag, method_name, **kwargs)

Cross validate sample value predictions for sample type sample_tag and readout data_tag_readout, using prediction method method_name.

Parameters

• data_tag_readout (str) – The key for self.readout.data where the Readout instance is stored.

• sample_tag (str) – The sample for which the gaussian process will be modeled according to the position in self.plate_layout.data. E.g. for positive controls “pos”

• method_name (str) – The prediction method. E.g. gp for Gaussian processes.

evaluate_well_value_prediction(data_predictions, data_tag_readout, sample_key=None)

Calculate mean squared prediction error.

ToDo: Debug. Better: REWRITE!

filter(value_data_type, value_data_tag, value_type=None, condition_data_type=None, condition_data_tag=None, condition=None, return_list=True)

Get list of values for defined wells of the data tagged with data_tag. If value_type is set, check if all values conform with value_type.

Parameters

• condition_data_type (str) – Reference to PlateData instance on which wells are filtered for the condition.

• condition_data_tag (str) – Data tag for condition_data_type

• condition (method) – The condition expressed as a method.

• value_data_type (str) – Reference to PlateData instance from which (for filtered wells) the values are retrieved.

• value_data_tag (str) – Data tag for value_data_type.

• value_type (str) – The type of the return values.

• return_list (bool) – Returns a flattened list of all values

Returns

(list of x), where x are of type value_type, if value_type is set.

..todo: rename method from filter to get_data

flatten_data(wells, values)

flatten_values(values)

flatten_wells(wells)

get_data_for_gaussian_process(data_tag_readout, sample_tags)

map_coordinates(coordinates_list)

preprocess(methodname, **kwargs)

randomize_values(data_tag_readout, data_tag_randomized_readout, randomized_samples='s', **kwargs)

Randomize the signal in a readout per plate and for a specific sample. The result of this method has only visualization purposes.

Parameters

• data_tag_readout (str) – The key for self.readout.data where the readouts are stored.

• data_tag_randomized_readout (str) – The key for self.readout.data where the randomized data will be stored.

• **kwargs –

subtract_readouts(data_tag_readout_minuend, data_tag_readout_subtrahend, data_tag_readout_difference, **kwargs)

un_flatten_data(y)

write(format, path=None, return_string=None, *args)

Serialize and write Plate instances.

Serialize Plate instance using the stated format.

Parameters

• format (str) – The output format: Currently only “pickle”.

• path (str) – Path to output file

Todo

Write checks for format and path.

hts.plate.plate.translate_coordinate_humanreadable(coordinate, pattern=None)

hts.plate.plate.translate_humanreadable_coordinate(humanreadable)

hts.plate.prediction module

Module contents